On Fri 23-06-17 07:05:37, Kees Cook wrote:
> On Fri, Jun 23, 2017 at 6:59 AM, Michal Hocko <mhocko@xxxxxxxxxx> wrote:
[...]
> >> --- a/fs/exec.c
> >> +++ b/fs/exec.c
> >> @@ -220,8 +220,18 @@ static struct page *get_arg_page(struct linux_binprm *bprm, unsigned long pos,
> >>
> >> if (write) {
> >> unsigned long size = bprm->vma->vm_end - bprm->vma->vm_start;
> >> + unsigned long ptr_size;
> >> struct rlimit *rlim;
> >>
> >> + /*
> >> + * Since the stack will hold pointers to the strings, we
> >> + * must account for them as well.
> >> + */
> >> + ptr_size = (bprm->argc + bprm->envc) * sizeof(void *);
> >> + if (ptr_size > ULONG_MAX - size)
> >> + goto fail;
> >> + size += ptr_size;
> >> +
> >> acct_arg_size(bprm, size / PAGE_SIZE);
> >
> > Doesn't this over account? I mean this gets called for partial arguments
> > as they fit into a page so a single argument can get into this function
> > multiple times AFAIU. I also do not understand why would you want to
> > account bprm->argc + bprm->envc pointers for each argument.
>
> Based on what I could understand in acct_arg_size(), this is called
> repeatedly with with the "current" size (it handles the difference
> between prior calls, see calls like acct_arg_size(bprm, 0)).
>
> The size calculation is the entire vma while each arg page is built,
> so each time we get here it's calculating how far it is currently
> (rather than each call being just the newly added size from the arg
> page). As a result, we need to always add the entire size of the
> pointers, so that on the last call to get_arg_page() we'll actually
> have the entire correct size.
Ohh, I forgot about this tricky part. The code just looks confusing
becauser we are mixing 2 things together here. This deserves a comment I
guess.
Other than that feel free to add
Acked-by: Michal Hocko <mhocko@xxxxxxxx>
Thanks!
--
Michal Hocko
SUSE Labs
